ChatGPT and Underwriting: Can We Chat?

The incentive to improve and innovate has been around for as long as the business world itself. The American engineer, Frederick Taylor, deemed that work is worthy of observation and analysis (1). In the early 20th century, Taylor pioneered the time study of work and workers, requiring the observer to use a stopwatch while standing next to the worker, timing task completion (2). In the early 1990s, business consultants Michael Hammer and James Champy coined the term “Reengineering”, launching a movement that aimed to upend the conventions of usual business practices, exhorting managers to take a fresh look at every function and task (3).

What is ChatGPT?

ChatGPT may become the “reengineering” of the 2020s. Poised to “change the world”, it uses advanced artificial intelligence (AI) that takes questions from users and produces human-like responses (4). GPT stands for “Generative Pre-Trained Transformer”, denoting the design and nature of the AI training. It is a reboot of artificial intelligence with much disruptive potential. The driving force behind ChatGPT appears to be its’ ability to optimize prompt engineering, using a question-based synthesis of natural language, learning and a higher degree of reasoning in responding to inquiries. There is great promise, but the debate is already in high gear. Useful or distracting? Helping or damaging? Already there are many questions, but few definitive answers at this point.

ChatGPT and Insurance Underwriting

What about ChatGPT as the potential fountainhead of all underwriting knowledge and wisdom? Is the underwriting manual of 2030 going to be a simple link to the latest Chat site, inputting analytics sourced from the e-application, calibrating as fast, if not faster than the fastest Google search?

We are not there yet. The ChatGPT site cautions current limitations including occasionally generating incorrect information, occasionally producing harmful instructions or biased content and lastly, limited knowledge of world and events after 2021 (5). The nature of risk selection requires a clear-eyed understanding of the uses and limitations of data and technology. It also requires leavening with human capital, the mix of experience and skill that builds the expertise necessary for successful underwriting (6). Simply put, challenging cases still require the human touch.

As an experiment,  ChatGPT was asked if it could evaluate an individual insurance risk. After some discussion of how risks are evaluated, the reply concluded with the following; “If you have questions about your insurance risk level, it’s best to consult with a licensed insurance agent who can evaluate your individual situation and provide personalized advice.” Pretty smart.

AI tools like ChatGPT will help accelerate underwriting transformation. Underwriters and their business partners will keep making sure that transformation is done the right way and for the right reasons.

  1. Frederik Winslow Taylor. N.D.
  2. Time and Motion Study. N.D.
  3. Magazine. Reengineering. February 9, 2020.
  4. Business Insider. Bill Gates calls ChatGPT ‘every bit as important as the PC’ or the internet. February 2, 2023.
  6. Cusick, Kelly, Ferris, Andy and Van Dalen, Britton. The rise of the exponential underwriter. February 24, 2021.

This article is provided by Know the Risk, an educational website that contains underwriting information for insurance professionals, available exclusively to Advisors affiliated with PPI (login required).